Pi Auto Media Type Detection for AiO A/A4 Copy and Standalone Photo Printing Functions
description
Transcript of Pi Auto Media Type Detection for AiO A/A4 Copy and Standalone Photo Printing Functions
© 2006 Hewlett-Packard Development Company, L.P.The information contained herein is subject to change without notice
Pi Auto Media Type Detection
for AiO A/A4 Copy and Standalone Photo Printing
Functions
Pi Project Team
March 20, 2007
2
Background Story• Customer selectable paper type is vulnerable• Customers don’t know what “Plain Paper or
Photo Paper” button was referring to: original media or output media
• No $ for Legolas sensor, Legolas sensor is not 100% correct.
• Legolas sensor was used only to confirm or reject plain or photo paper type in the copy function
3
Background Story• Need to remove a button to allow space for the
power button (moved from front faceplate), and enable a simpler UI for cutomers
Fullhouse UI Pi UI
4
Application/specific cases
• Standalone copy or
photo printing from card slots
and
• Only on large (A/A4) size media; small media (4x6 or L size depending local setting) defaults to photo media
5
Usage ModelMilhouse Use Model
Product Life 3 Years
Number of Printed Pages Scans Only Scan and Copy1 Year 2 Years 3 Years 1 Year 2 Years 3 Years 1 Year 2 Years 3 Years
Avg. 1428 2856 4284 46 91 137 487 974 1461Mean 1031 2062 3093 30 60 91 325 650 97590th Percentile 2857 5714 8571 95 189 284 1021 2042 306395th Percentile 3718 7435 11153 126 251 377 1353 2706 405998th Percentile 5434 10868 16303 203 406 608 2185 4370 6555
65% of all printed pages are prints from PC35% of all printed pages are copies
Ink Usage (cc's per month) cc's per year
Black Color Black ColorAvg. 3.6 1.6 43.2 19.2Mean 2.7 1.1 32.4 13.290th Percentile 6.6 3.4 79.2 40.895th Percentile 8.5 4.4 102 52.898th Percentile 11.7 5.7 140.4 68.4
Photo ink usage is expected to be only 0.06cc's per month.
Media UsagePlain 95%Inkjet 1%Photo 4% 25% of these are 4x6 media
Photo Media Pages1 year 2 years 3 years
Average 46 92 13890th Percentile 106 212 31895th Percentile 230 460 69098th Percentile 478 956 1434
Print Modes - PC PrintDraft 7%Normal 82%Best 11%
Print Modes - CopyDraft 7%Normal 88%Best 5%
Pi usage model not available, this is to give a sense of the media detection
• Is applied to no more than 35% of the total printed pages• Will mostly need to detect 3% (=4% x 75%) photo media over the
machine life time
Photo card printing information not available, estimated to be a small percentage of everyday printing
6
Technical Model
Calibration MoveMove paper motor at 5ips and
20ips and record avg pwm
Paper Load move
Record PWM data through
MECH_REC
Width sensor?
Return default type for small
media
Call PWM_media_detect to get paper
type
NEW!
Calibrate raw pwm using avg
pwm values
Calculate stats like sum, mean, std_dev
for different time zones
Run stats through a statistical model to find
out cardstock vs noncardstock
Return Plain Paper type
Run a third model to determine Plain
vs Photo
Return Photo paper type
Return Plain Paper type
Run a second model to determine
EvDPh
Return Photo paper type
7
Math Model
• One time factory calibration• No-Load Calibration before each job• Back EMF Correction• Discriminant Analysis: probability
result• Compare result to threshold value to
determine if Plain or Photo
8
Problem:
Plain and Photo Paper PWM Pick Profiles can be Indistinguishable Due to Variations in:
• Motor temperature• Motor torque constants• Part dimensions and friction• Ambient temperature and humidity• Power supply voltage• Life (wear)•
Proposed Solution:
Perform one-time factory no-load calibration to measure room temp baseline
Perform a No-Load PWM Calibration before every Job
Calibration
9
RAW No-Load Data vs. Motor Temp.
50
100
150
200
250
300
350
5 7 9 11 13 15 17 19 21 23 25
No-Load Shaft Speed (ips)
PW
M
26°C
43°C
52°C
64°C
No-Load Data w/ Back-EMF Correction vs. Motor Temp.
50
100
150
200
250
300
350
5 7 9 11 13 15 17 19 21 23 25
No-Load Shaft Speed (ips) (Torque)
Co
rrec
ted
PW
M
26°C
43°C
52°C
64°C
No-Load PWM Calibration Details:1. Average PWM is measured before each pick at a no-load shaft speeds of 8 ips and 20 ips.
2. The PWM values are corrected for back-EMF to approximate the effective no-load system torque constant.
As a Result:The differences between each calibrated pick profileshould now be mostly due to paper differences only.
3. The resultant linear no-load PWM-to-Torque relationship is used to convert all raw pick profile data into the equivalent calibrated no-load torque.
10
-170
-160
-150
-140
-130
-120
-110
-100
1 34 67 100 133 166 199 232 265 298 331 364 397 430 463 496 529 562 595
Encoder Counts
PW
M
RAW Pick Profile Data:
— Plain Paper,Motor = 28°C
— Plain Paper,Motor = 64°C
— Photo Paper,Motor = 28°C
— Photo Paper,Motor = 64°C
Plain @ 64°C Overlaps Photo @ 28°C
Calibration Example Using Hot Motor Data:
11
-12
-11
-10
-9
-8
-7
-6
-5
1 37 73 109 145 181 217 253 289 325 361 397 433 469 505 541 577
Encoder Counts
Eq
uiv
ale
nt
No
-Lo
ad
Sh
aft
Sp
ee
d(E
qu
iva
len
t N
o-L
oa
d T
orq
ue
)
CALIBRATED Pick Profile Data:
— Plain Paper,Motor = 28°C
— Plain Paper,Motor = 64°C
— Photo Paper,Motor = 28°C
— Photo Paper,Motor = 64°C
The Temperature Variation is Calibrated Out,and Plain is Separated from Photo
Cont: Calibration Example Using Hot Motor Data:
12
1 21 41 61 81 101 121 141 161 181 201 221 241 261 281 301 321 341 361 381 401 421 441 461 481 501 521 541 561 581
Encoder Counts
Zone 7
Zone 6
Zone 5
Zone 2 Zone 3 Zone 4Zone 1
Discriminant Analysis:The following calculations are performed on the calibrated profile datato the select zones (1, 2, 4, 7):
• Mean• Standard Deviation
13
Discriminant AnalysisSample Equation:
14
15
Latest Classification Results
16
Gen II Improvements for Rho:1. determine 3 thresholds--one favoring plain, one optimized, one favoring photo.
2. choose one of the 3 different thresholds based on the customer's selection of quality setting.
3. choose one of the 3 different thresholds based on the customer's selection of size.
4. choose one of the 3 different thresholds based on whether the customer is making a copy, using the "reprints" mode, or printing from a card.
5. choose one of the 3 different thresholds based on whether the customer is making a color copy or a b/w copy.
6. fix the problem of random picks at plain paper pick speed.
7. increase the constant pick speed from the current 6? ips to a more stable 8 ips.
8. detect multi-picks.
9. detect paper-out condition (as differentiated from a no-pick).
10. add a plain paper pick calibration to EOL.
11. add a plain paper pick calibration customer walk-through during initial setup (OOBE), or simply use the paper pick from the first pen calibration as the plain paper calibration profile.
17
Gen II Improvements for Rho cont.:1. add a plain paper pick re-calibration accessible to the customer from the
service menu, or simply re-calibrated every time a pen alignment page is printed (by assuming plain paper, and ignore if it's clearly not plain paper).
2. tweak the thresholds adaptively based on customer usage--for example, if customer makes a draft mode b/w copy, they are most likely using plain paper, so use that pick's sqdistdiff to adjust the thresholds accordingly.
3. communicate the detected paper type to the UI, so that it can be displayed in the cgd during the copy job (see Rho's latest flash UI demo).
4. tweak the thresholds adaptively when "grey" type media is detected based on customer's response to a question during the copy job underway screen on the cgd, "glossy (matte) paper detected, correct?" and have the buttons mapped to "yes" and "no".
5. tweak the thresholds based on customer's region.
6. detect 4x6 vs 5x7 for small media.
7. use customer's size selection history of 4x6 and 5x7 to adaptively optimize/tweak the 4x6 vs 5x7 threshold.
18
Pi Team Credits:
Brad Short 858.655.6918Fahad Ahmad 858.655.3230Ardi Othman 011.65.6727.1568Tushar Radke 011.65.6727.2906Jeff Borra 858.655.4417